False Positive and Negative Rates of Pancreatic Endoscopic Ultrasound-Guided Cytologic Diagnosis: One Institution's Experience.
Kirsten Woolf, Zachary Sletten, Donna Russell, Thomas Bonfiglio, Zhongren Zhou. University of Rochester Medical Center, NY
Background: Primary diagnosis of pancreatic tumors is primarily based on endoscopic ultrasound (EUS) guided fine needle aspiration biopsy (FNA) and bile duct brushing (BDB). Diagnosis of malignancy is often complicated due to sampling difficulty and concomitant reactive changes associated with inflammation and atypia. Both false positive and false negative results lead to significant clinical consequences. For the purpose of quality control, we have reviewed our pancreatic cytology cases for the last six years.
Design: 733 pancreatic FNA biopsy cases including solid and cystic lesions were identified in our departmental files from 2004-2010. 264 pancreas resections were performed at our institution during this time period, of which 134 had prior cytologic diagnosis in our files (101 FNA and 33 BDB). Using resection diagnosis and clinical information, we compared both histological and cytological diagnoses to determine the false positive and false negative rate in our EUS-FNA diagnoses. All slides were independently reviewed by 3 pathologists (KW, TB, ZZ).
Results: 733 FNA cases were divided into 3 categories: positive or suspicious (290, 40%); negative or atypical (403, 55%) and unsatisfactory (40, 5%). Of the 134 cytologic cases (FNA and BDB) that had corresponding resections, there were 70 (52%) positive cases, 60 (45%) negative or atypical cases, and 4 (3%) unsatisfactory cases. Of those, 101 (75%) FNA cases included 58 (57%) positive, 39 (39%) negative, and 4 (4%) unsatisfactory diagnoses. There were 43 (32%) cases among the 134 cytology cases that had a diagnostic discrepancy (FNA, 20; BDB, 23). Among the 20 FNA cases, the majority of these represent a false negative diagnosis (19) with only one false positive case. The total false positive rate when including all positive FNA cases is 0.3% (1/290); while the total false negative rate is 5% (19/403). Among false negative FNA cases, only two cases were found to be negative due to interpretative error (2/403, 0.5%), while the remaining cases were negative due to sampling. Six cases (6/23) with interpretive error were found in the discrepant BDB cases.
Conclusions: EUS-FNA diagnosis for pancreatic lesions has very few false positives (0.3%) and false negatives (5%). The major cause of false negatives by FNA is due to sampling (17/19). Interpretative error of false negatives by cytopathologists is very low (0.5%). BDB was more likely to lead to a false interpretative error (BDB vs FNA, 6/23 vs 2/19) due to marked inflammation and the presence of ductal stents.
Tuesday, March 1, 2011 9:30 AM
Poster Session III # 61, Tuesday Morning